Project Summary/Abstract Clinical trials remain the gold standard for evidence-based assessment of investigational treatments and are essential to the development of desperately needed therapies for Alzheimer's disease (AD). AD trials, including those sponsored by NIH and industry, now enroll patients with dementia and mild cognitive impairment (MCI). The cognitive status of these populations necessitates enrollment of two individuals, a participant and a study partner. The quality of AD trial data is inherently tied to the study partner's ability to provide complete, accurate, and consistent assessments of patient outcomes. If study partners are not qualified to fill their role or are replaced, patients may dropout or their data may be at risk for inaccuracy. This proposal examines whether particular study partners disproportionately contribute to trial variance and potential bias due to conditional confounding. We hypothesize that study partner characteristics are associated with variance and bias in trials. We have assembled data from actual AD and MCI trials, enabling studies to examine our hypothesis with the greatest generalizability possible. Specifically, we will examine whether study partner characteristics are associated with trial recruitment, retention, and data integrity outcomes. We also will examine the impact of informant replacement on trial data. We will perform modeling experiments that investigate the implications of altering trial enrollment patterns and protocol decisions on trial data precision and accuracy. Finally, we will develop software that incorporates our results and enables trialists to better plan their studies. This project will yield a greater understanding of study partner impact on trial data integrity and will enable improved design and conduct to reduce missing data and increase statistical power and data validity.!